Aim: Test the effect of sematic & acoustic similarity on the learning & recalling of word sequences in LTM using control to prevent rehearsal between presentation & testing to minimise STM effects.
Procedure:78 ppts used in independent measures. Lab based experiment. Ppts male and female selected from applied psychology research unit.
Condition A: 21 ppts, list of 10 acoustically similar words.
Condition B: 20 ppts, list of 10 acoustically dissimilar words.
Condition C: 16 ppts, list of 10 semantically similar adjectives.
Condition D: 21 ppts, list of 10 semantically dissimilar adjectives.
4 trials – each presented with list of words individually, appearing for 3 seconds, distraction of intervening digit sequence for 8 seconds, test and had 1 minute to write 10 words down (repeated 4 times).
15 minute self-paced digit copying task after the 4 trials.
Ppts finally given a surprise recall task of the 10 word list in correct order; testing of the word recalled visible in random order. (sequence of words being tested)
Results: A 77%, B 68%, C 67%, D 86%.
30% SD between semantically similar & dissimilar words. 9% SD between Acoustically similar & dissimilar.
Similar list (A) harder to learn than control (B) at first but by trial 4 it was reversed. No sight of forgetting between end trial 4 & final recall group. No sig difference between similar & dissimilar. (Acoustic similar hard to learn than dissimilar, but got better by end)
Semantic: Trial 1 little difference in learning words, but by trial 4 slower learning for similar semantic group (C) , recall scores sig higher for control (D). Neither list showed any forgetting between trial 4 & recall task – semantically similar more poorly recalled.
Conclusion: Learning of words impaired by semantic similarity. Encoding in LTM is extensively semantic. Encoding in STM largely acoustic while LTM is largely semantic.
| Strengths | Weaknesses |
| High R – conducted in a highly controlled lab with a standardized procedure of each condition having the same words, 4 trials & surprise recall – can replicate for consistencies.
High R – underwood & goads results showed how LTM affected by semantic similarity – if others can get similar results can be considered reliable. Application – shows LTM encodes & recalls better for semantically dissimilar words – helps to apply to revision techniques where you don’t study same topics on the same day. High Internal V – use of a control group which controls Ev & allows for manipulation of the IV, ruling out STM effects – can established a cause & effect relation. |
Low G – uses a poor sample of independent groups from uni research panel (78 ppts) – cannot generalise to whole population & biased of psychological knowledge.
Low task V – involved learning lists of monosyllabic words under strict conditions – the study doesn’t represent everyday memory process and lacks mundane realism. Low eco V – study took place in an artificial environment with strict control of Ev – meaning that variables which effect normal memory aren’t present so memory may not be natural behaviour. Reductionist – memory which is a complex human process is reduced down to recall of the order of 10 words – doesn’t replicate real life use or memory.
|